Computational chemistry as atomistic magnifying glass for biological systems

Studying biological macromolecules such as proteins in their entirety on the atomistic level is a huge experimental challenge. Prof. Dr. Christine Peter from the University of Konstanz is therefore working on the development of special computer models that are able to calculate and visualize the complex interaction of atoms. She is using these models to study the chemical basis of complex biological phenomena such as protein folding and the stability of viral capsids.

In order to study the physicochemical foundations of phenomena as complex as the self-aggregation of proteins and the biomineralization of bone, all atoms and the forces acting on them must be taken into account. However, in practice, classical laboratory experiments are unable to do this.

Christine Peter, professor of theoretical and computational chemistry at the University of Konstanz, is therefore adopting a completely different approach to the study of the chemical foundations of biological structures and processes. Peter and her research group are developing computer simulations that enable them to look inside systems. "We are using computers for our experiments in order to obtain a microscopic picture of the structures under investigation," says Peter, talking about her research design.

Simulation methods - investigating, testing and verifying

While laboratory experiments are usually only used to study an average of a huge number of atoms or molecules rather than individual atoms or molecules, computer simulations enable scientists to specifically alter and study individual parameters on the high-resolution level. Computational methods therefore provide scientists with information about the driving forces behind experimental or real-life phenomena. "It is like using an atomistic magnifying glass to look at atoms and molecules in action," says Prof. Peter.

However, before the scientists are able to carry out computer simulations, they need to design a model that contains all the necessary parameters, in particular the chemical interactions between atoms and molecules as well as reference values from experimental data. "For example, a water model would need to include density and dielectricity constant data," says Peter. The model would also have to include data derived from quantum-chemical calculations. The researchers need to spend a lot of time investigating, testing and verifying the required parameters. "We need to come up with a reliable set of model parameters for a system under investigation before we can get started with the real work," says Prof. Peter.

Simulations connect experiments with theory

First, each model needs to be validated using knowledge obtained in experiments carried out on the system under investigation. In a subsequent step, computer simulations can also help to explain experimental data and provide microscopic interpretation. "Experimental measurements do usually not provide direct information about the position of atoms or the structure of large molecules, one rather obtains indirect information such as signals or spectra," says Peter. "Computer simulations give us an idea about the positions, structures and movements of the atoms and molecules involved in a particular experiment. Knowledge relating to the connection between the two enables us to calculate the expected experimental signal."

The results envisaged in theory and the actual measurements can then be compared. "If they match, this then confirms the simulation model as well as the microscopic interpretation of the experiment," says the chemist. The computer simulations can be used for reproducing and explaining known data and experiments as well as for proposing new experiments and making prognoses about unknown factors. "The simulation then adds another piece to the jigsaw puzzle of the overall understanding of a system or phenomenon," explains Prof. Peter. As this new piece of the puzzle represented by the data cannot always be directly verified in the laboratory, it has to be examined theoretically within the framework of the system to find out whether it fits convincingly into the data obtained with laboratory experiments and simulations and whether it provides a coherent overall picture of the system under investigation.

Simulations at several levels of resolution - from all-atom to coarse-grained descriptions

Christine Peter's research not only involves models for small molecules. She also studies complex biological systems such as protein folding and aggregation. Single-atom level models can be used to simulate the folding of relatively short peptide chains on a time scale of up to a few microseconds.

However, simulating the folding of larger proteins and the formation of protein aggregates requires time scales of milliseconds and even seconds. This can only be achieved on the coarse-grained level. Investigations on the mesocopic level thus also enable the researchers to look at groups of atoms that are summarized as units, rather than just single atoms. "With these models we can simulate larger systems and longer time scales whilst retaining the computing power," says Prof. Peter, referring to some of the advantages of the models used. The challenge lies in using hierarchical simulation approaches - i.e. multiscale simulation methods - that link simulations at several levels of resolution in order to produce an overall picture. Coarse-grained models are produced from models of higher resolution by leaving out increasing numbers of elements.

"When we produce coarse-grained models from higher resolution models, we need to be aware that the level where 'chemistry is taking place', i.e. the level of atomic interactions, is not taken into account, which means that the model can no longer represent all the details," says Prof. Peter, highlighting the limitations of the approach. Using a multiscale simulation approach, the researchers need to keep in mind which issues a simplified model can be used for and whether such a model is in fact suitable for making predictions. If these aspects are taken into account, the researchers can then access new scales and deal with biologically relevant questions.

"For example, we use our models to study the assembly of large multiprotein complexes in cells and for elucidating the reasons for the mechanical stability of viral capsids," says Prof. Peter. The researchers have already found out that viral capsids are locally reinforced with so-called beta barrels that enable viruses to better withstand pressure. "The previous theoretical model did not take this viral property into account. So our research has been able to close a gap in the theory," says Prof. Peter.